• Publications
  • Influence
On the structure of hidden Markov models
A Note on Metric Properties for Some Divergence Measures: The Gaussian Case
TLDR
This paper proposes a modication for the KL divergence and the Bhattacharyya distance, for multivariate Gaussian densities, that transforms the two measures into distance metrics and shows how these metric axioms impact the unfolding process of manifold learning algorithms.
Pareto discriminant analysis
TLDR
Experimental results on synthetic data, several image data sets and data sets from the UCI repository show positive and encouraging results in favor of PARDA when compared with standard and state-of-the-art multiclass extensions of LDA.
Designing a Metric for the Difference between Gaussian Densities
TLDR
Based on the proposed metric, a symmetric and positive semi-definite kernel between Gaussian densities is introduced and illustrated in two settings: a supervised problem, where the user learns a low-dimensional projection that maximizes the distance between Gaussians, and an unsupervised problem on spectral clustering.
An a Priori Exponential Tail Bound for k-Folds Cross-Validation
TLDR
A priori generalization bounds developed in terms of cross-validation estimates and the stability of learners are considered and an exponential Efron-Stein type tail inequality is derived for the concentration of a general function of n independent random variables.
Classification of time-series data using a generative/discriminative hybrid
TLDR
A general generative/discriminative hybrid that uses HMMs to map the variable length time-series data into a fixed p-dimensional vector that can be easily classified using any discriminative model is proposed.
What Is the Distance Between Objects in a Data Set?: A Brief Review of Distance and Similarity Measures for Data Analysis
TLDR
Digitally recorded data have become another critical natural resource in the current research environment and marked the beginning of another unstoppable activity that is intimately related to digitally stored data: extracting knowledge and information from such data.
A generative-discriminative hybrid for sequential data classification [image classification example]
TLDR
A general generative-discriminative framework that uses HMMs to map the variable length sequential data into a fixed size P-dimensional vector (likelihood score) that can be easily classified using any discriminative model is proposed.
The Minimum Volume Ellipsoid Metric
TLDR
An unsupervised "local learning" algorithm for learning a metric in the input space yielding the MVE metric (MVEM) which showed promising and competitive results when compared with state of the art metrics in the literature.
...
1
2
3
...